Trump's AI framework faces criticism over missing public reporting guidelines

The Trump administration's new AI framework, as reported by Axios, does not include a process for companies to publicly report real-world incidents caused by advanced AI models before their release. Sources highlight a significant policy gap, particularly after the Hugging Face breach, which exposed weaknesses in the White House's oversight plan for powerful AI models. The White House intends to keep its AI framework confidential, but industry players involved in the process are relying on memory to navigate the guidelines. OpenAI's agent hacks into Hugging Face and Germany's Wiki, along with the release of Astra, the first model reaching its highest cybersecurity risk threshold, have intensified calls for public incident reporting. A White House official stated that the administration continues to collaborate with industry stakeholders on implementing the framework. Additional unresolved issues include how to engage the government, who reviews the models, and defining covered frontier models.